Method and apparatus for determining a document suitability for server-based optical character recognition (OCR) processing

Abstract

There is disclosed a method of analyzing a digital image of a document (to determine, as example, a document suitability for server-based OCR processing) in a computer system that includes a user electronic device (for acquiring or storing a digital image of a document) connectable to a server (for executing the server-based OCR processing of the digital image to create a recognized-text document). The method is executable by the user electronic device and comprises: acquiring the digital image of the document; analyzing an OCR quality parameter associated with a compressed digital image to be created from the digital image using a compression algorithm and a compression parameter; in response to the OCR quality parameter being above or equal to a pre-determined threshold: transmitting the compressed digital image to the server. Optionally, the method further comprises compressing the digital image using the compression algorithm and the compression parameter to create the compressed digital image before transmission thereof.

What is claimed is: 
     
       1. A method of analyzing a digital image of a document, the method executable
 by a user electronic device, the method comprising: 
 acquiring, by the user electronic device, the digital image of the document; 
 analyzing, by a classifier executed by the user electronic device, an optical character recognition (OCR) quality parameter associated with a test digital image, the test digital image created by compressing and decompressing the digital image using a compression algorithm and a compression parameter; and 
 in response to the OCR quality parameter being above or equal to a pre-determined threshold: 
 transmitting a compressed digital image to a server for the server to execute the server-based OCR processing, the compressed digital image created from the digital image using the compression algorithm and the compression parameter. 
 
     
     
       2. The method of  claim 1 , wherein prior to the transmitting, the method further comprises compressing the digital image using the compression algorithm and the compression parameter to create the compressed digital image. 
     
     
       3. The method of  claim 2 , in response to the OCR quality parameter being below the pre-determined threshold not executing:
 compressing the digital image using the compression algorithm and the compression parameter to create the compressed digital image; 
 transmitting the compressed digital image to the server for the server to execute the server-based OCR processing. 
 
     
     
       4. The method of  claim 3 , wherein the method further comprises, in response to the OCR quality parameter being below the pre-determined threshold, causing user electronic device to acquire another digital image of the document. 
     
     
       5. The method of  claim 1 , wherein the analzying the OCR quality parameter comprises:
 compressing the digital image using the compression algorithm to create a test compressed digital image, the compression algorithm using the compression parameter; 
 de-compressing the test compressed digital image to create a decompressed test digital image; 
 applying the classifier to the decompressed test digital image to determine the OCR quality parameter, the OCR quality parameter being indicative of whether the decompressed test digital image having been compressed using the compression parameter is suitable to render an acceptable outcome by the server-based OCR processing. 
 
     
     
       6. The method of  claim 1 , wherein the analyzing the OCR quality parameter comprises:
 executing a first compressing the digital image using the compression algorithm to create a first test compressed digital image, the compression algorithm using a first compression parameter; 
 de-compressing the first test compressed digital image to create a first test decompressed digital image; 
 executing a second compressing the digital image using the compression algorithm to create a second test compressed digital image, the compression algorithm using a second compression parameter; 
 de-compressing the second compressed digital image to create a second test decompressed digital image; 
 applying the classifier to the first decompressed test digital image to determine a first OCR quality parameter and to the second decompressed digital image to determine a second OCR quality parameter; 
 selecting one of the first compression parameter and the second compression parameter based on an associated one of the first OCR quality parameter and the second OCR quality parameter being indicative of a higher suitability of the respective one of the first decompressed test digital image and the second decompressed test digital image for rendering an acceptable outcome by the server-based OCR processing; and wherein the compressing the digital image using the compression algorithm to render the compressed digital image is executed by using the selected one of the first compression parameter and the second compression parameter. 
 
     
     
       7. The method of  claim 1 , wherein compression algorithm is implemented as a codec selected from at least one of: JPEG and JPEG 2000. 
     
     
       8. The method of  claim 7 , wherein the analyzing the OCR quality parameter associated with the test digital image is further based on compression data obtained from the codec. 
     
     
       9. The method of  claim 1 , wherein the classifier is implemented as a machine learning algorithm. 
     
     
       10. The method of  claim 9 , further comprising training the machine learning algorithm. 
     
     
       11. The method of  claim 9 , further comprising selecting those text blocks with the smallest font and wherein the determining the OCR quality parameter is executed on the selected text blocks with the smallest font. 
     
     
       12. The method of  claim 1 , wherein the analyzing the OCR quality parameter comprises:
 defining four contour images; 
 splitting each of the four contour images into non-overlapping blocks; 
 determining, for each of the four contour images, an average contrast; 
 generating binary version of the image in order to define contour pixels; 
 determining sharp pixels of each of the blocks; 
 determining the number of contour pixels for each of the blocks; 
 determining text blocks; and 
 determining the OCR quality parameter based on the text blocks. 
 
     
     
       13. A system for analyzing a digital image of a document, the system comprising:
 a memory; and 
 a processor, operatively coupled to the memory, being configured to: 
 acquire, the digital image of the document; 
 analyze, using a classifier executed by the processor, an OCR quality parameter associated with a test digital image, the test digital image created by compressing and decompressing the digital image using a compression algorithm and a compression parameter; and 
 in response to the OCR quality parameter being above or equal to a pre-determined threshold: 
 transmit a compressed digital image to a server for the server to execute the server-based OCR processing, the compressed digital image created from the digital image using the compression algorithm and the compression parameter. 
 
     
     
       14. The system of  claim 13 , wherein the processor is further configured to compress the digital image using the compression algorithm and the compression parameter to create the compressed digital image. 
     
     
       15. The system of  claim 14 , the processor being further configured, in response to the OCR quality parameter being below the pre-determined threshold not to execute:
 compressing the digital image using the compression algorithm and the compression parameter to create the compressed digital image; 
 transmitting the compressed digital image to the server for the server to execute the server-based OCR processing. 
 
     
     
       16. The system of  claim 15 , wherein the processor is further configured, in response to the OCR quality parameter being below the pre-determined threshold, to cause user electronic device to acquire another digital image of the document. 
     
     
       17. The system of  claim 13 , wherein to analyze the OCR quality parameter, the processor is configured to:
 compress the digital image using the compression algorithm to create a test compressed digital image, the compression algorithm using the compression parameter; 
 de-compress the test compressed digital image to create a decompressed test digital image; 
 apply the classifier to the decompressed test digital image to determine the OCR quality parameter, the OCR quality parameter being indicative of whether the decompressed test digital image having been compressed using the compression parameter is suitable to render an acceptable outcome by the server-based OCR processing. 
 
     
     
       18. The system of  claim 13 , wherein to analyze the OCR quality parameter, the processor is configured to:
 execute a first compressing the digital image using the compression algorithm to create a first test compressed digital image, the compression algorithm using a first compression parameter; 
 de-compress the first test compressed digital image to create a first test decompressed digital image; 
 execute a second compressing the digital image using the compression algorithm to create a second test compressed digital image, the compression algorithm using a second compression parameter; 
 de-compress the second compressed digital image to create a second test decompressed digital image; 
 apply the classifier to the first decompressed test digital image to determine a first OCR quality parameter and to the second decompressed digital image to determine a second OCR quality parameter; 
 select one of the first compression parameter and the second compression parameter based on an associated one of the first OCR quality parameter and the second OCR quality parameter being indicative of a higher suitability of the respective one of the first decompressed test digital image and the second decompressed test digital image for rendering an acceptable outcome by the server-based OCR processing; and wherein the compressing the digital image using the compression algorithm to create the compressed digital image is executed by using the selected one of the first compression parameter and the second compression parameter. 
 
     
     
       19. The system of  claim 13 , wherein the compression algorithm is implemented as a codec selected from at least one of: JPEG and JPEG 2000. 
     
     
       20. The system of  claim 19 , wherein to analyze the OCR quality parameter associated with the test digital image, the processor is configured to consider compression data obtained from the codec. 
     
     
       21. The system of  claim 13 , wherein the classifier is implemented as a machine learning algorithm. 
     
     
       22. The system of  claim 21 , wherein the processor is further configured to execute training the machine learning algorithm. 
     
     
       23. The system of  claim 13 , wherein to analyze the OCR quality parameter, the processor is configured to:
 define four contour images; 
 split each of the four contour images into non-overlapping blocks; 
 determine, for each of the fourth contour images, an average contrast; generate binary version of the image in order to define contour pixels; determine sharp pixels of each of the blocks; determine the number of contour pixels for each of the blocks; determine text blocks; 
 determine the OCR quality parameter based on the text blocks. 
 
     
     
       24. The system of  claim 23 , the processor being further configured to select those text blocks with the smallest font and wherein the determining the OCR quality parameter is executed on the selected text blocks with the smallest font. 
     
     
       25. Non-transitory computer-readable media for analyzing a digital image of a document, the non-transitory computer-readable media having recorded thereon instructions that when executed by one or more computer processors, perform operations comprising:
 acquiring, by a user electronic device, the digital image of the document; 
 analyzing, by a classifier executed by the user electronic device, an OCR quality parameter associated with a test digital image, the test digital image created by compressing and decompressing the digital image using a compression algorithm and a compression parameter; and 
 in response to the OCR quality parameter being above or equal to a pre-determined threshold: 
 transmitting a compressed digital image to a server for the server to execute the server-based OCR processing, the compressed digital image created from the digital image using the compression algorithm and the compression parameter. 
 
     
     
       26. A method of analyzing a digital image of a document, the method executable by a user electronic device, the method comprising:
 acquiring, by the user electronic device, the digital image of the document; 
 executing a classifier executed by the user electronic device, the classifier having been trained to predict an OCR quality parameter associated with the digital image based on primary characteristics associated with the digital image without having to compress the digital image; and 
 in response to the OCR quality parameter being above or equal to a pre-determined threshold: 
 transmitting a compressed digital image to a server for the server to execute the server-based OCR processing, the compressed digital image created using a compression algorithm and a compression parameter. 
 
     
     
       27. The method of  claim 26 , wherein the primary characteristics include at least one of: a font size, number of characters, number of characters per page, brightness of the image, or contrast level of the image. 
     
     
       28. The method of  claim 26 , wherein the classifier is implemented as a machine learning algorithm. 
     
     
       29. The method of  claim 28 , further comprising training the machine learning algorithm.